The Microdeformation Fabric of Amphibole-Rich Peridotite in the Southern Mariana Trench and Its Influence on Seismic Anisotropy

Abstract

Olivine, the most abundant mineral in the upper mantle, exhibits elastic anisotropy. Understanding the seismic anisotropy and flow patterns in the upper mantle hinges on the crystallographic preferred orientation (CPO) of olivine. Similarly, hydrous minerals, which also display elastic anisotropy, play a crucial role in explaining seismic anisotropy in numerous subduction zones. High-temperature and -pressure simple shear experiments reveal that the CPO of amphibole can lead to significant seismic anisotropy. In this study, peridotite samples originating from the southern end of the Mariana Trench, commonly containing amphibole, were analyzed. The microdeformation fabric and seismic anisotropy were examined. The results indicate a weak fabric strength in olivine, yet identifiable deformation fabrics of A/D, D, and AG were observed. Various dislocation structures suggest that olivine experiences complex deformation across various temperatures. Not only can the original slip system transform, but the melt/fluid resulting from melting also has a substantial impact on the peridotite. Deformation precedes the melt/rock interaction, resulting in a strong melt/rock reaction under near-static conditions. Furthermore, the modal content of amphibole significantly alters the seismic anisotropy of peridotite. An increase in amphibole content (types I, III, and IV) enhances seismic anisotropy, particularly for type I amphibole. Notably, the presence of type I fabric amphibole promotes the Vs1 polarization direction parallel to the trench in subduction zones, a phenomenon observed in other subduction zones. Therefore, when considering mantle peridotite regions rich in amphibole, the impact of amphibole on seismic anisotropy must be accounted for.

1. Introduction

Olivine, the most abundant mineral in the upper mantle, exhibits strong intrinsic seismic anisotropy (Mainprice et al., 2000 and references within), which can result in seismic anisotropy at the scale of the whole rock depending on the preferred orientation (also called fabric) of its fast [100] axis, and hence gives unique seismic anisotropy patterns depending on the fabric type [1,2,3,4]. Understanding the observed seismic anisotropy of the rocks, such as rapid S-wave polarization anisotropy, radial anisotropy, P-wave, and Vp/Vs azimuth anisotropy, is crucial for understanding geodynamic processes in the upper mantle [5,6,7]. Despite extensive work to elucidate the genesis of these diverse olivine fabrics [7,8,9,10,11,12,13,14,15,16,17,18], the mechanisms of their formation remain poorly constrained [12,13,14,15,16,17,18]. A simple model of olivine fabric has been established in the last decades, presenting the fabric types in a three-dimensional space encompassing temperature, stress, and content, which holds significant implications for inferring the temporal and spatial distribution of olivine fabrics and their impact on seismic anisotropy in subduction zones [12,13,14,15,16,17,18,19,20,21]. These olivine CPOs are commonly categorized into five fabric types, A, B, C, D, and E, which have been established based on the results of experimental studies and another olivine fabric type, previously known as the axial-[010] or [010]-fiber pattern named AG type, following on from the five fabric types of Karato [22]. Alternatively, Précigout and Almqvist (2014) proposed a distinct distribution model for olivine fabric in the mantle wedge region [23]. Despite variations among these models regarding the distribution of olivine fabric in the mantle wedge, they share a common assumption that the forearc mantle primarily comprises a single type of olivine fabric. However, this hypothesis may only hold to a limited extent. Indeed, coexistence of distinct olivine fabrics [22] as well as predominance of A-/D-type fabrics [24,25] have been frequently observed in natural peridotites. None of the models previously cited can fully explain these observations, regardless of whether it originates from forearc or back-arc environments [26,27,28,29,30,31,32,33,34,35,36]. The olivine A/D-type fabric is formed at a dry lower lithospheric forearc mantle postdating its intense partial melting in a juvenile subduction zone [24].

Moreover, it is crucial to consider the significant presence of water-bearing minerals, such as serpentine, chlorite, and amphibole, in subduction zones, including the forearc mantle [37,38,39,40,41,42,43]. The influence of these aqueous mineral phases on seismic anisotropy remains poorly understood. Experimental studies on the simple shear CPO of amphibole under 1 GPa of pressure have demonstrated that the amphibole CPO can induce significant seismic anisotropy in the lower crust and shallow subduction zones, with up to 14.6% P-wave velocity anisotropy (AVp) and 12.1% maximum S-wave velocity anisotropy (AVs) [43,44,45,46,47,48]. Hence, a crystallographic preferred orientation (CPO) in amphibole may lead to significant seismic anisotropy, as it has been shown in experiments under high temperature and pressure conditions [43]. Therefore, amphibole may play a pivotal role in explaining the seismic anisotropy observed in subduction zones, especially in hydrated peridotites within the mantle wedge [41].

At subduction zones, the splitting of several phases (local S, source-side S, and vertically traveling SKS waves) is measured in order to discriminate among different anisotropy sources, like those potentially located in the supra-slab, slab, and sub-slab mantles [49]. A strong trench-parallel anisotropy is commonly found in the sub-wedge mantle (i.e., slab + sub-slab mantle), with δt possibly proportional to the trench migration rate [7]. Data on the seismic wave anisotropy near Guam indicate that the polarization direction of the fast waves (S1) shows that, to the north of Guam, the fast wave direction is parallel to the direction of the Pacific Plate’s motion, while in the southwest of Guam, it mainly displays a fast wave direction parallel to the island arc [50].

Our understanding of the microstructures in olivine in the Mariana Trench is still very limited. Harigane et al. (2013) examined the deformation fabric of harzburgite in the southern landward slope of the Izu–Ogasawara Trench, and suggest that it preserves an early-stage mantle structure from the subduction zone [20]. Michibayashi et al. (2007) conducted preliminary studies on peridotites in the southern Mariana Trench, revealing significant fabric variations among samples, attributed to mantle melting and wedge complexities [27]. Oya et al. (2022) focused on peridotites in the southernmost Mariana Trench, indicating associations between grain size, fabric characteristics, and interactions with melts beneath the lithospheric mantle [51]. Similar findings were observed east of the Challenger Deep, suggesting peridotites may originate from large-scale ductile shear zones. These investigations provide valuable insights into the geological evolution of trench regions. However, the fabric of amphibole in peridotites from the Mariana Trench and its impact on seismic wave anisotropy remain unclear.

This study delves into the olivine and amphibole present in the amphibole-rich peridotite of the southern Mariana Trench. It examines the relationship between CPO type (including deformation mechanisms) and seismic anisotropy in naturally deformed peridotites. Furthermore, it discusses the influence of the amphibole fabric type and modal content on seismic anisotropy. This research offers a novel perspective to enhance our understanding of the deformation fabric of peridotite in the forearc mantle of current subduction zones and the influence of amphibole on seismic anisotropy.

2. Geological Setting and Sample Collection

The Mariana Trench is the location where the Pacific Plate subducts beneath the Philippine Sea Plate, a process that commenced approximately 43 million years ago [52]. This subduction activity has resulted in the formation of a series of residual and active island arcs within the Philippine Sea Plate. The Western Mariana Ridge and the Kyushu–Palau Ridge lie to the west of the active Mariana Trough. To the west of the trench, the Parece Vela Basin and the active Mariana Island Arc are situated in succession (Figure 1). Notably, the Izu–Bonin–Mariana (IBM) arc, with a width of approximately 200 km, stands apart from other arcs due to its absence of an accretionary wedge. This uniqueness is attributed to the arc system’s isolation from terrigenous sediments, leading to limited availability of pelagic sediments and pyroclastic materials. The nearly exposed IBM arc front offers an excellent opportunity for rock sampling.

Understanding the initiation mechanism of subduction is a key point for models of plate tectonics. Although studies on the initiation of subduction have proposed both spontaneous and induced mechanisms, the detailed process of subduction initiation remains controversial [52,53]. In the past decade, people’s understanding of the onset of subduction has mainly focused on systematic chronological and geochemical studies of the IBM arc front. The spontaneous subduction of IBM began at ~52.5 Ma and occurred along the boundary of the conversion fault between the original Philippine Plate and the Pacific Plate [54]. This is due to the convergence of the buoyant lithosphere of the original Philippine Plate with the denser lithosphere of the Pacific Plate [55]. In the early stages of subduction, the pre arc basalt was formed by decompression and melting of the depleted asthenosphere mantle, forming the current pre arc [56]. The association between sheet-like veins and pillow-shaped basalt in the pre arc crust indicates that the pre arc basalt was formed during the overlying plate expansion process between 52.5 and 48 Ma [55]. The continuous subsidence of the lithosphere and the infiltration of plate-derived fluids led to further fluid-assisted partial melting of the depleted peridotite mantle, resulting in the production of glassy andesite around 48–44 Ma. The seafloor expansion also affected the backarc region, forming the Amami Sankaku Basin in the west of Kyushu Basin around 49–48 Ma [57,58]. Finally, normal tholeiitic basalt and calc alkaline arc magmatism occurred around 44 Ma (7–8 Ma after the start of subduction), and the location of magmatic activity migrated westward from the trench [55]. The duration of pre arc magmatic activity is less than 10 Ma [55,58].

Samples obtained by trawling on the landward slope of the Izu–Bonion–Mariana Trench mainly consist of igneous rocks, including peridotite, basalt, and island arc tholeiitic basalt [59,60,61]. Similar sequences of basalt and island arc tholeiitic basalt were also obtained during the Deep Sea Drilling Project (DSDP) and the Ocean Drilling Program (ODP) [62,63]. These rocks formed during the initial development stage of the Eocene Izu–Ogasawara–Mariana subduction zone [64]. The peridotite samples investigated from the southern Mariana forearc have undergone substantial serpentinization. Moreover, these samples display signs of seafloor alteration, evidenced by a delicate Mn-oxide layer coating their surfaces. Certain peridotite samples exhibit discernible color gradations from their cores to peripheries, transitioning from a brownish-green hue at the center to a progressively yellow-brown shade toward the edges.

Our peridotite samples were collected during the Hakuho R/V KH03-3 expedition conducted by the University of Tokyo’s Ocean Research Institute. They were obtained using trawling methods from the landward slope of the southern Mariana Trench (Figure 1b). Sample locations include KH03-03-D08: 143°26.43′ E, 11°48.40′ N, at a depth of 3500 m, and KH98-1-D3: 143°26.45′ E, 11°43.50′ N, at a depth of 5700 m (Figure 1b). Due to severe alteration and weathering on the seafloor, mineral grains are prone to scatter during thin section preparation. Therefore, prior to thin sectioning, a consolidation process was performed by putting the sample in boiling glue and soaking it. The samples were sequentially polished using 600 grit, 9.5 µm, and 3 µm sandpaper with Al₂O₃ polishing powder to thin the sample surface, achieving a thickness of approximately 30 µm. Subsequently, mechanical polishing was conducted using polishing fluids of 1 µm and 0.05 µm with a cloth, followed by chemical-mechanical polishing using a 0.05 µm suspended SiO₂ polishing fluid.

3. Materials and Methods

3.1. EBSD Data Acquisition and Processing

The sample texture was determined based on the microstructures of the olivine, amphibole, and spinel components. To investigate the CPOs of olivine and amphibole, thin sections were prepared in the X-Z plane, with ‘X’ representing the lineation and ‘Z’ being perpendicular to the foliation (Figure 2a). EBSD (electron backscatter diffraction) analysis was conducted at the Scanning Electron Microscopy Laboratory of the State Key Laboratory of Geological Processes and Mineral Resources at China University of Geosciences (Wuhan). Utilizing an FEI Quanta 450 field emission scanning electron microscope equipped with an Oxford Instruments HKL Nordlys II electron backscatter diffractometer, high-energy electron beams were generated for diffraction analysis. This setup allowed for synchronized control of the sample stage, image acquisition, and EBSD testing. The specific test parameters were: acceleration voltage of 20 kV, spot size of 6, working distance of 23 mm, step size of 20 μm, and sample tilt of 70°. The Nordlys II detector automatically collected the electron backscattering patterns (EBSPs), which were then processed using Aztec software. For EBSD data processing, the MTEX software package developed on MATLAB was used (http://mtex-toolbox.github.io, accessed on 25 March 2024) [65,66]. Compared to traditional Channel 5 software, MTEX offered more efficient and batch processing capabilities.

The MTEX software package can be used to create custom visualizations and calculations based on individual requirements. These include: the alignment factor (AF)—this factor quantifies the alignment of grains or particles within the material relative to a specific direction. It is useful for assessing anisotropy in material properties that may result from preferred grain orientations; grain size (d)—in this context, grain size refers to the equivalent circular diameter, which is the diameter of a circle having the same area as the grain. This measurement method ensures consistency and comparability of grain size across different samples and conditions; aspect ratio (AR)—this parameter measures the shape of the grains, defined as the ratio of the longest axis to the shortest axis of a grain. A higher aspect ratio indicates more elongated grains, which can influence the mechanical properties of the material; shape factor (SF)—the shape factor indicates how closely the shape of a grain approximates a perfect circle (in 2D) or sphere (in 3D). It is typically calculated as the ratio of the perimeter of the grain to the perimeter of a circle with the same area, with values closer to 1 indicating more circular or spherical grains; grain orientation spread (GOS)—GOS measures the variability of crystallographic orientations within individual grains. A higher GOS value suggests a greater degree of misorientation within the grain, which can affect the material’s strength and ductility; and misorientation to mean (M2M)—this parameter quantifies the average misorientation of the crystal lattice within a grain relative to the grain’s average orientation. It is useful for understanding internal strain and dislocation density within grains.

This study employs pole figures to represent the crystallographic orientation of peridotite minerals. These pole figures are projected onto the lower hemisphere, with a half-width set to 10. The fabric strength of the minerals is quantitatively assessed using the J-index and M-index [67,68]. Before any calculations, the EBSD maps were cleaned by eliminating isolated pixel points and filling isolated gaps. The pole figures were plotted using one-orientation per grain.

3.2. Calculation of Seismic Properties

The seismic wave velocity and anisotropy of the whole rock were computed based on the measured CPOs for olivine and amphibole, the relative phase content of the samples, and the single-crystal elastic moduli and density from the literature. The calculation then used the Voigt–Reuss–Hill approximation as the averaging method [8]. The analysis was performed using the MATLAB-based MTEX toolbox (https://mtex-toolbox.github.io, accessed on 25 March 2024) [8,65]. The P-wave velocity anisotropy (AVp) was determined using the following formula:

$$AVp=200\times \left(V{p}_{\max }-V{p}_{\min }\right)/\left(V{p}_{\max }+V{p}_{\min }\right)$$

(1)

And S-wave velocity anisotropy (AVs) was calculated as follows:

$$AVs=200\times \left(V{s}_1-V{s}_2\right)/\left(V{s}_1+V{s}_2\right)$$

(2)

where Vs₁ is the velocity of the fast shear wave, and Vs₂ is the velocity of the slow shear wave.

The polarization of the fast wave and the delay time associated with shear waves propagating in anisotropic rocks provide quantitative descriptions of anisotropy. The delay time between the fast and slow waves is influenced by the degree of anisotropy and the thickness of the anisotropic layer [69,70]. The single-crystal elastic constants and density data for olivine and amphibole were obtained from Brown and Abramson (2016) and Abramson et al. (1997), respectively [4,47].

3.3. Microstructure of Dislocation

To observe the dislocation structure of olivine, we decorated the dislocations in the crystals using the following method. The single-side polished sample was heated in a CNC high-temperature furnace, with a temperature increase of 200 °C every 15 min, until it reached 900 °C for 1 h. This caused the surface of the sample to oxidize. Once the power was turned off, the sample was cooled naturally to room temperature and then processed into standard sheets using conventional methods.

FeO components are oxidized along dislocations in minerals, and light brown or brown hematite (or magnetite) precipitates on the dislocation lines when the temperature drops, thus decorating and setting off the dislocation configuration characteristics. The dislocation observation and statistics were conducted using scanning electron microscopy (SEM) at the Laboratory of Environmental Scanning Electron Microscopy, China University of Geosciences (Wuhan). The dislocation decoration was performed on samples KH03-3-D8-003, KH98-3-D3-SP9, KH98-3-D3-SP7, KH03-3-D8-136, and KH03-3-D8-156.

4. Results

4.1. Microstructure

The peridotite from the southern Mariana forearc, studied in this research, typically exhibits a porphyroclastic texture, with olivine as the porphyroclast and a matrix composed of olivine, amphibole, and spinel. Some olivines display wavy extinction, and cracks are often present. Larger olivine crystals are frequently interconnected and altered. Spinel is predominantly round to sub-round, diamond-shaped, dark brown, and opaque, often observed in larger grain sizes, and occasionally worm-like (Figure 2). The primary mineral composition of the peridotite in the southern Mariana Arc comprises ol (58.1–98.6 vol.%), amp (0–41.0 vol.%), and sp (0.1–5.7 vol.%) (

Table 1. Mineral composition and structure by EBSD analysis.

Sample Number	Mineral Assemblage (vol%)			Microstructure
	ol	amp	sp
KH03-3-D8-003	93.43	5.81	0.76	Coarse
KH98-3-D3-SP9	91.81	8.08	0.10	Coarse
KH98-3-D3-SP7	90.66	9.07	0.28	Coarse
KH03-3-D8-002	90.63	9.06	0.31	Coarse
KH03-3-D8-156	58.14	41.34	0.52	Porphyroclastic
KH03-3-D8-122	98.59	0.00	1.41	Porphyroclastic
KH03-3-D8-136	92.18	7.00	0.82	Porphyroclastic
KH03-3-D8-124	91.18	8.68	0.15	Porphyroclastic
KH03-3-D8-011	81.24	17.47	1.29	Porphyroclastic
KH03-3-D8-101	84.14	15.17	0.69	Fine
KH03-3-D8-102	78.07	19.93	1.99	Fine
KH03-3-D8-120	93.99	0.26	5.74	Fine

). The alteration minerals, including serpentine, chlorite, and talc, account for approximately 10–40 vol.%.

Ohara et al. (1998) categorized the peridotites collected from the southern Mariana Trench into three types: anhydrous (type A), hydrous (type H), and intermediate (type I) [45]. As our samples are pure peridotite without pyroxene and generally contain amphibole, they fall under the H-type peridotite classification by Ohara et al. (1998) [45]. Our samples are composed of olivine, amphibole (0.3–1.5 mm), and spinel (1–3 mm).

The olivine crystals in our samples can be broadly categorized into three grain size ranges: >1 mm (up to 5 mm), 0.1–0.4 mm, and <0.1 mm. In samples such as KH03-3-D8-136, KH98-3-D3-SP9, KH03-3-D8-003, and KH03-3-D8-011, olivine with a grain size > 1 mm dominates. These olivine crystals exhibit curved contact boundaries, and small size olivine crystals are often observed near the larger ones. The larger olivine crystals also display well-developed wavy extinction. In other samples, olivine with a grain size ranging from 0.1 to 0.4 mm predominates, and these crystals often develop a triple-point structure. Occasionally, the olivine appears elongated. Olivine with a grain size < 0.1 mm is primarily found in the microfracture zones commonly developed in the rock. It appears as intermittent veins and is often filled with late-stage alteration minerals such as amphibole, chlorite, serpentine, and talc. Amphibole is typically columnar or needle-like and often exhibits a preferred orientation aligned with the lineation. Microscopically, some samples exhibit relatively undeformed coarse-grained or fragmentary structures. In other samples, there is a lower abundance of residual phenocrysts or a finer-grained structure, with the matrix composed of aligned olivine, amphibole, and spinel particles along cracks, forming a shape preferred orientation (SPO). These samples tend to have a higher amphibole content. It is worth noting that different structures can coexist within the same rock, and the distribution of amphibole content can be uneven even within the same sample.

4.2. CPOs of Minerals

The CPOs of olivine and amphibole as well as the microstructural parameters were measured using EBSD in our samples from the southern Mariana Trench, and are reported in Figure 3 and

Table 2. Microstructure parameters of peridotite minerals in the southern Mariana Trench.

Sample Number	Olivine								Amphibole
	J-Index	M-Index	AF	d (μm)	AR	SF	GOS	M2M (°)	J-index	M-index	d (μm)	AR	SF	GOS	M2M (°)
KH03-3-D8-003	1.92	0.07	53.04	168.57	1.75	1.27	1	1.56	1.42	0.02	73.31	1.71	1.21	1.46	1.72
KH98-3-D3-SP9	1.77	0.07	8.14	107.48	1.56	1.2	0.86	2.47	1.16	0.01	80.98	1.59	1.2	1.18	2.2
KH03-3-D8-011	1.29	0.03	16.08	102.4	2.55	1.29	0.83	1.46	2.51	0.07	67.25	2.02	1.23	1.04	2.27
KH98-3-D3-SP7	1.31	0.03	2.78	123.66	1.55	1.21	0.73	6.24	1.23	0.01	88.02	1.56	1.19	0.92	1.71
KH03-3-D8-002	3.66	0.14	43.81	154.03	1.63	1.21	1.09	2.93	1.81	0.03	43.49	1.78	1.24	1.25	3.23
KH03-3-D8-156	1.15	0.02	29.98	122.78	1.49	1.17	0.77	1.61	6.13	0.22	101.6	1.67	1.21	1.23	2.42
KH03-3-D8-122	2.21	0.1	34.69	127.96	1.59	1.2	0.91	3.01
KH03-3-D8-136	1.32	0.03	12.86	114.77	1.49	1.18	0.98	2.04	2.3	0.07	111.68	1.51	1.16	1.42	2.35
KH03-3-D8-124	1.3	0.03	29.11	157.4	1.51	1.18	0.75	1.07	3.54	0.13	89.77	1.53	1.16	0.71	1.12
KH03-3-D8-101	1.45	0.05	18.55	67.28	1.54	1.18	0.76	1.31	2.66	0.07	57.38	1.59	1.18	1.33	1.99
KH03-3-D8-102	1.26	0.02	11.73	120.64	1.48	1.18	0.86	1.48	2.21	0.06	108.39	1.5	1.16	1.22	2.76
KH03-3-D8-120	2.56	0.11	20.88	133.58	1.37	1.12	0.85	0.98
KH03-3-D8-104	1.75	0.06	28.36	77.63	1.5	1.13	0.71	0.79

Note: AF, shape preference index; d, grain size; AR, axis rate; SF, shape index; GOS, intracrystalline orientation distribution; M2M, the orientation difference relative to the average orientation. Fabric strength is calculated based on each mineral as a point (PPG).

, respectively. Although the fabric strength is generally weak (J = 1.15–3.66), a clear crystallographic preferred orientation is observed in the olivine phase of most of the studied samples. In the olivine samples studied, we observed three distinct types of crystallographic preferred orientation (CPO). The first CPO is observed in samples KH03-3-D8-011, KH03-3-D8-101, and KH03-3-D8-102, where the [100] axis forms dense clusters of points aligned parallel to the lineation. This corresponds to an A/D-type fabric [24,25]. The second CPO appears in samples KH03-3-D8-003, KH03-3-D8-104, and KH98-3-D3-SP9, characterized by an extremely dense alignment of the [100] axis points parallel to the lineation, while the [010] and [001] axes form a ring belt oriented perpendicular to the lineation, indicative of a pronounced D-type fabric [22]. The third CPO, observed in sample KH03-3-D8-136 and KH03-3-D8-156, exhibits an AG-type fabric [22,71].

In our samples, the olivine J-index ranges from 1.15 to 3.66 and the M-index ranges from 0.02 to 0.14 (Table 2). Figure 4 shows that olivine from all samples exhibits varying distribution patterns for the rotation axis with 2–10° disorientation. The strong density of the rotation axis is concentrated between the [001] and [010] axes. The presence of multiple extreme densities for the rotational axis distribution within a single sample indicates the coexistence of multiple slip systems within the same sample or even within individual olivine crystals. A-type CPO is generally attributed to dislocation slip on the (010) [100] slip system, although other slip systems such as (010) [100] and (001) [100] are sometimes also active (albeit with varying proportions) [72]. The superposition of these different slip systems results in the formation of CPO patterns with relatively weak fabric strength.

Recently, the amphibole CPOs were classified into four types and are proposed to be closely related to temperature, differential stress, and shear strain [41,45]. The amphibole crystallographic orientation manifests primarily in four patterns: dense clustering of the [001] axes, with annular or random distribution of the [100] and [010] axes (Ⅳ type); dense clustering of the [100] axes with annular or random distribution of the [001] and [010] axes(III type); weak fabric with random orientation; and strong density of the [100], [001], and [010] axes in the case of KH03-3-D8-156 (I type) (Figure 3). The oblique angle between the direction of the maximum CPOs of olivine [100] and amphibole [001] is estimated to be between 15° and 32° (Figure 3). This angle is comparable to the oblique angles observed between the direction of the maximum CPOs of olivine [100] and pyroxene [001] in naturally deformed peridotites [73]. This similarity suggests that the amphibole may have inherited the structure of the primary pyroxene during the diagenetic process.

Thus, based on the CPO pattern (Figure 3) and the rotation axis inverse pole figure (Figure 4), the olivine fabric in the southern Mariana Trench exhibits characteristics of type A, type A/D (intermediate between type A and type D), and type AG.

Figure 4. Distribution characteristics of inversion of rotation axis of olivine adjusting 2–10° orientation difference in peridotites in the southern Mariana Trench. The inset in the lower right corner shows the relationship between the olivine low-angle (2–10°) misorientation axes inverse pole figure and the slip systems. (Modified according to De Kloe, 2001 [74]). The specific judgment method is as follows: (1) When the rotation axis of the low-angle particle boundary is the [100] axis, olivine can only slip and deform through [001] (010); (2) When the rotation axis is the [001] axis, olivine can only be deformed by [100] (010) sliding; (3) When the rotation axis is the [010] axis, if the low-angle boundary is a tilt wall, olivine will be deformed by [001] (100) or [100] (100) slip, and if the low-angle boundary is a twist wall, olivine will pass through [001] (010) and [100]. (4) When the rotation axis is {0kl}, olivine can only be deformed by [100] {0kl} sliding, excluding the [010] and [001] rotation axes already discussed.

Figure 4. Distribution characteristics of inversion of rotation axis of olivine adjusting 2–10° orientation difference in peridotites in the southern Mariana Trench. The inset in the lower right corner shows the relationship between the olivine low-angle (2–10°) misorientation axes inverse pole figure and the slip systems. (Modified according to De Kloe, 2001 [74]). The specific judgment method is as follows: (1) When the rotation axis of the low-angle particle boundary is the [100] axis, olivine can only slip and deform through [001] (010); (2) When the rotation axis is the [001] axis, olivine can only be deformed by [100] (010) sliding; (3) When the rotation axis is the [010] axis, if the low-angle boundary is a tilt wall, olivine will be deformed by [001] (100) or [100] (100) slip, and if the low-angle boundary is a twist wall, olivine will pass through [001] (010) and [100]. (4) When the rotation axis is {0kl}, olivine can only be deformed by [100] {0kl} sliding, excluding the [010] and [001] rotation axes already discussed.

4.3. Seismic Anisotropy

Interestingly, the pattern of seismic anisotropy is roughly similar across all samples (Figure 5), with the maximum and minimum P-wave velocities parallel and perpendicular to the lineation, respectively, the maximum and minimum velocities of fast shear waves parallel and perpendicular to the foliation plane, respectively, and the maximum anisotropy of the S wave when propagating along the normal to the foliation plane. The strength of the anisotropy, however, varies widely: maximum P-wave anisotropy (AVp) ranges from 4.5% to 12.5%, maximum S-wave anisotropy (AVs) is between 3% and 8.2%, and fast wave (S1) anisotropy (AVs1) ranges from 1.7% to 7.1%.

Sample KH03-3-D8-156, in particular, contains an extremely high amount of amphibole compared to the other samples, reaching 41%. This KH03-3-D8-156 sample exhibits the strongest fabric strength of amphibole (J-index 6.13) among all samples, but the weakest olivine fabric strength (J-index 1.15). The EBSD map in Figure 6 shows that the sample is actually divided into two regions, a region relatively rich in amphibole (Amp-rich) and a region with less amphibole (Amp-poor), which are separated by a fissure in the middle of the sample. Different olivine structures coexist in the sample. In the Amp-poor region, the olivine exhibits a porphyroclastic texture, with a large number of fine-grained olivine crystals distributed around large porphyroclasts. By contrast, in the Amp-rich region, the olivine has a fine-grained structure. The orientation of the amphibole is pronounced, with well-developed small cracks, and most of the amphibole grows along these cracks. A detailed study of the two regions revealed that the olivine fabric in the Amp-poor region (J-index 1.41) is stronger than in the Amp-rich region (J-index 1.30). This may suggest that the presence of amphibole reduces the strength of the olivine fabric, which sparked our interest in studying the effect of amphibole crystallographic preferred orientation (CPO) on the seismic anisotropy of peridotite.

4.4. Dislocation Microstructure of Olivine

Scanning electron microscopy (SEM) was utilized to investigate the microstructure of dislocations in olivine. As depicted in Figure 7, across all studied samples, following oxidation of pendants, dislocations primarily appear as white lines and points, with less frequent occurrences as curves or rings. Furthermore, dislocations are not uniformly distributed among grains or within individual crystals. These results indicate that the olivine in this region exhibits a complete range of dislocation types and abundant forms, reflecting the plastic deformation experienced by the upper mantle, primarily dominated by high-temperature dislocation creep. The primary dislocation configurations observed are: (1) free dislocation—a single, randomly distributed dislocation within the crystal; (2) dislocation row and dislocation wall—a dislocation wall, a low-energy configuration, is composed of numerous dislocations arranged in a specific pattern to form a dislocation row. This arrangement marks a surface defect in the crystal. The neatly aligned dislocations within the wall occasionally intersect, maintaining a roughly equal spacing. This type of inclined wall, composed primarily of edge dislocations, is a typical structure formed during the steady-state flow of minerals at elevated temperatures [75]; (3) dislocation arch bend and dislocation ring—when a dislocation is generated, both ends of the dislocation become fixed during plastic flow. If the dislocation remains in a state of high temperature with very low strain rate, it will climb and slide, eventually forming a dislocation ring. The arch bend of a dislocation also arises from dislocation movement under steady-state conditions, typically initiated above 800 °C [76]; (4) dislocation grid—this structure arises from two sets of closely parallel, horizontally aligned spiral dislocations, often forming a typical rectangular or diamond shape; and (5) subgrains—these polygonal substructures are formed by the climbing and cross-slip of free dislocations during recovery, with small angular deflections in the lattice orientation constrained by the dislocation wall.

5. Discussion

5.1. Olivine Slip Systems and Deformation Mechanism

Significant research has focused on understanding the plastic deformation mechanisms in olivine, which are critical for interpreting seismic anisotropy and mechanical behavior in the upper mantle. Studies such as Jung et al. (2009) and Tielke et al. (2019) have extensively investigated dislocation slip in olivine, revealing insights into how pressure and water content influence these mechanisms [31,77]. Additionally, diffusion creep, another pivotal deformation mechanism, has been thoroughly examined in works like Mei et al. (2000), highlighting the role of water in facilitating this process. Further exploration of Dis-GBS and other mechanisms could provide a more comprehensive understanding of olivine deformation under various geophysical conditions. The microstructures of a sample are indicative of the deformation mechanisms that have occurred in the sample, and these deformation mechanisms can themselves be attributed to large-scale geological processes. For example, the deformation of coarse-grained olivine. The deformation of coarse-grained olivine (~2–4 mm) typically indicates high-temperature deformation within the solid or super-solid regime, associated with asthenosphere flow beneath spreading centers (1200–1250 °C); it is thus a ‘fossil’ microstructure [78]. Conversely, low-temperature deformation (1000–1100 °C) under the lithosphere leads to the formation of porphyroclastic structures, accompanied by varying degrees of recrystallization and the development of fine-grained (<1 mm) layers [79]. For instance, observations from sample KH98-3-D3-SP7 revealed a correlation between olivine fabric type, strength, and grain size (Figure 8). As grain size decreases, the orientation of olivine crystal axes becomes weaker, and randomness increases. Similar findings were obtained from statistical analyses of representative samples KH03-3-D8-156 and KH03-3-D8-003, indicating that the presence of fine-grained olivine results in reduced fabric strength (Figure 8).

Experimental studies have consistently demonstrated that the (010) [100] slip system is the most active at elevated temperatures within the {0kl} [100] slip system family [72,73,74,75,76,77,78,79,80,81,82]. Coarse-grained olivine (>0.8 mm) from our samples exhibits a dominant (010) [100] slip system. However, in some samples, such as KH98-3-D3-SP7 (Figure 8), coarse-grained olivine has also been observed to exhibit a (001) [100] dominant slip system. This variation may come from an early mantle flow change or late-stage addition of aqueous melts, which activate the (001) [100] slip system. The observation of changes in slip direction within the same crystal, as dislocation walls bend by ~40° (Figure 7d), supports the multiple deformation steps hypothesis. The coexistence of different types of dislocations within the same crystal further suggests that olivine crystals undergo dislocation slip at varying temperatures, with free dislocations forming at relatively low temperatures and dislocation arches and rings developing at higher temperatures within the same crystal (Figure 7).

Harigane et al. (2013) studied the deformation fabric of gabbro in the southern land of the Izu–Bonin Trench [20]. They believe that these gabbro peridotites preserve the mantle structure formed in the early stage of the subduction zone and provide information of the initial stage of Pacific Plate subduction under the Philippine Sea Plate. Olivine fabric is likely to be A type or D type formed under the condition of low flow stress and no water in the diving infancy, and it will change into C type or E type with the addition of water and the increase in water-bearing fluid derived from the plate.

Michibayashi et al. (2007) first made a preliminary study on the deformation fabric of peridotite in the south of the Mariana Trench, and the results showed that there were great differences in olivine fabric among different samples. (010) [100] and (010) [001] fabrics (type A and type B) were observed in coarse-grained samples, but no clear fabrics were observed in fine-grained samples [27]. This greatly resembles the microstructures observed in our samples. Peridotite samples obtained from the trench slope on the south side of the Mariana Trench represent a wide range of refractory residues after mantle melting [83] and also represent a complex component of the mantle wedge, which Michbayashi et al. (2007) believes is caused by the vertical and lateral thinning of the overlying lithosphere, which leads to the asthenosphere expanding upward and toward the trench [27]. Dehydration of the subduction plate may lead to hydration of the shallow asthenosphere uplift area [84] and then lead to partial melting of mantle wedge, which is consistent with the existence of island arc volcano in the Mariana system west of Guam [84]. Therefore, the variable micro-fabric features in our samples well reflect the complex tectonic evolution in the southern part of the Mariana Trench.

The trench in the region extends through the forearc to the backarc of the Mariana arc system, and this structural pattern also largely causes differences in the olivine fabric characteristics of the region in the E–W oriented geographic space. For example, recent research by Oya et al. (2022) on peridotite in the southernmost part of the Mariana Trench, at the junction of the Challenger Deep and the Yapu Trench, has shown that the olivine fabric varies with grain size, with coarse-grained peridotite being (010) [100] (type A) and fine-grained peridotite being {0kl} [100] (type D), while samples with different grain sizes exhibit various inconspicuous patterns [51]. Fine-grained peridotite (D-type fabric) may undergo deformation under relatively high flow stress, indicating the presence of a ductile shear zone in the deep 139° E ridge. These mineralogical and geochemical features may be attributed to the interaction between melts and rocks in the relatively shallow lithospheric mantle, which is more similar to peridotite in the Parece Vela Basin [51] rather than Mariana arc peridotite.

Coincidentally, similar research results also appear in the east of the Challenger Deep, near the north–south strike of the West Santa Rosa Bank Fault (WSRBF), one left-lateral fault zone, and the West Guam fault (WGF). The crystal orientation of olivine in peridotite shows a typical [100] (010) (type A) model, and the fabric strength gradually increases from coarse-grained rocks to mylonitic texture [28]. Therefore, these peridotites may originate from large-scale ductile shear zones developed in the mantle of the overlying lithosphere, may be in anhydrous conditions, and may be located above the tearing of subduction plates. The peridotite is now exposed along the north–south strike fault zone of West Santa Rosa Beach. The composition of spinel is consistent with that of the back-arc peridotite in the Mariana Trough, indicating that the typical back-arc mantle is lacking in this area, and it is replaced by the back-arc basin mantle exposed along the West Santa Rosa Beach fault [28].

The complex tectonic evolution of peridotite in the southern Mariana Trench is reflected in the strong intracrystalline deformation and weak crystallographic preferred orientation (CPO) strength of olivine. This evolution has resulted in a series of changes to the slip systems originally present in the mantle. The finer grain size observed in olivine indicates higher flow stress [84,85], suggesting that fine-grained peridotites may have formed in shallower, lower-temperature environments compared to porphyritic peridotites. Alternatively, different microstructures may arise from varying degrees of deformation [29]. The widespread fractures and cleavage observed in peridotite samples from the southern Mariana Trench, along with the common occurrence of fine-grained olivine, may be linked to the north–south ductile shear zone of the West Santa Rosa Bank Fault (WSRBF) [28].

On top of these complex dislocation slip deformations, grain boundary sliding (GBS) can also be a possible deformation mechanism for olivine in our samples. The small olivine grains may have deformed primarily by GBS, while the CPO is a remnant of a ‘fossil’ CPO from sea floor spreading [86,87]. GBS is a key deformation mechanism in fine-grained materials, particularly effective under conditions of high temperature and differential stress. Through this mechanism, grains can slide past each other, accommodating large strains without significant intracrystalline deformation. Especially in mantle wedge areas or above subducting plates, fine-grained olivine and peridotite may exhibit GBS as the main mode of stress adaptation [86,87]. This deformation mechanism could dominate the current deformation process, while preserving the characteristics of the crystallographic preferred orientation (CPO) formed during past tectonic events.

5.2. Influence of Melt Percolation or Melt/Rock Interaction on Deformation

Studies have demonstrated that peridotites from the southern Mariana Trench have undergone intense partial melting, likely due to a slow melt extraction rate or significant porosity, allowing extended interaction time between the melt and peridotites [84,88]. Olivine exhibits a pronounced CPO and weak fabric strength. The high dislocation density and diverse dislocation types suggest complex deformation within the mantle across various temperatures. However, some olivine crystals in certain samples display inhomogeneous dislocation density, with notable regions of low dislocation density (Figure 6). The absence of pyroxene in our peridotite suggests the formation of new olivine crystals during pyroxene melting, indicating that late-stage melt flow altered the fabric of primary olivine. Furthermore, the presence of numerous fine olivine particles with randomly distributed crystal axes indicates that annealing has somewhat reduced the fabric strength of olivine. To discern the order of melting/rock interaction and deformation, a comparison was made between the fabric characteristics of olivine with varying grain sizes within the same peridotite. The results (Figure 8) revealed that fine olivine grains, particularly those smaller than 0.3 mm, exhibit random crystal axis distribution and lack a defined fabric.

The AG type (also known as axial 010 or “A-C switch”) typically forms in the presence of compression [72] or melt [14,71]. Recent studies on natural peridotite sample microstructure reinforced the idea that melt penetration is crucial in the transformation of the original A-type fabric. Specifically, as the melt/rock ratio increases, the melt penetration process systematically transforms the olivine crystal CPO from the initial A type to AG type [89]. This transformation is evident in sample KH03-3-D8-156, which exhibits an AG fabric and a corresponding high amphibole modal content.

This strong melt/rock reaction occurs in a near-static environment during subduction, where the infiltrating melt disrupts the connected network of olivine, increasing porosity. This disruption promotes sliding and rotation of the olivine grains under compaction, leading to the formation of new, rotating subgrains that transform into new crystals. The isogranular structure of certain particles and the emergence of distinct low dislocation density regions during crystal growth jointly contribute to the development of a weak and nearly random fabric [73]. Consequently, the deformation of this portion of the olivine can be attributed to the static melt/rock reaction.

5.3. Influence of Amphibole CPO on Seismic Anisotropy

Natural rock samples have also highlighted the importance of amphibole CPO in explaining seismic anisotropy in the crust [46]. However, despite the presence of up to ~15%–40% amphibole in the mantle wedge [90], there is a paucity of research examining how the amphibole CPO alters the seismic properties of water-bearing mantle rocks [91]. Subsequently, this section delves into the influence of amphibole on seismic anisotropy in amphibole-rich dunites.

In our peridotite samples, amphibole exhibits various CPO types. Specifically, type I CPO is observed in samples KH03-3-D8-002, KH03-3-D8-003, and KH03-3-D8-156, type III CPO in sample KH03-3-D8-102, and type IV CPO in samples KH03-3-D8-011, KH03-3-D8-101, KH03-3-D8-124, and KH03-3-D8-136 (Figure 3).

Experimental studies on amphibole deformation at 1 GPa and temperatures ranging from 480–700 °C have shown that amphibole CPO formation is significantly influenced by differential stress and temperature [91]. Type I amphibole CPO forms under relatively low temperatures and varying stress conditions, while type IV amphibole CPO develops under high shear strain [41,43]. These findings suggest that peridotites in the southern Mariana Trench have experienced low-temperature and high-shear deformation, consistent with the observed ductile shear zone in their microstructure.

The amphibole content in our peridotite samples is highly variable, ranging from 0 to 41%, with a particularly significant concentration in sample KH03-3-D8-156. The analysis of this sample revealed that the fabric strength of olivine is weaker in amphibole-rich areas compared to those with lower amphibole content. An analysis of seismic anisotropy in two distinct regions of sample KH03-3-D8-156, which vary in amphibole content (Figure 6), revealed that the directional patterns of anisotropy (i.e., the orientations of the fast and slow wave directions shown in Figure 5) exhibit minimal variation between the amphibole-poor and amphibole-rich regions. The strength of the anisotropy parameters, however, is lower in the Amp-poor region than in the Amp-rich region, with a difference of ~38% in AVp and ~54% in AVs. There are two potential sources to explain the difference in anisotropy between these regions: the presence of a small-grain population in the Amp-poor region that may decrease the CPO (and so the seismic anisotropy), and the presence of amphibole in the Amp-rich that may strengthen the CPO. In the following, we present seismic simulations in order to test if the hypothesis where the excessive presence of oriented amphibole can alone explain the difference in seismic anisotropy in these sample regions. The modal content of type I CPO amphibole in sample KH03-3-D8-156 significantly enhances the seismic anisotropy of peridotite (Model 1). Moreover, we used the experimental CPO measured in representative samples: KH03-3-D8-102 exhibiting type III CPO amphibole (olivine fabric type A) (Model 2) and KH03-3-D8-124 displaying type IV CPO amphibole (olivine fabric type D) (Model 3). We tested amphibole-to-olivine ratios of 0:100, 10:90, 20:80, 30:70, and 40:60 in MTEX and estimated their seismic anisotropy in the upper subduction zone. The findings are presented in Figure 9, Figure 10 and Figure 11.

The results reveal that, under a hypothetical scenario where no amphibole is present—i.e., when the samples are entirely composed of olivine—the seismic anisotropy of Model 1 is lower than that of Model 2 and Model 3. For all models, the increase in amphibole content is accompanied by an increase in the seismic anisotropy. Notably, Model 1, containing type I amphibole, exhibits the fastest growth in anisotropy. The AVs1 value particularly undergoes a drastic transformation, surpassing that of Models 2 and 3 with only a minimal increase in amphibole content (Figure 12). These findings suggest that the seismic anisotropy of peridotite is amplified by an increase in the content of all three CPO fabric types (types I, III, and IV), with type I amphibole exhibiting the most pronounced effect. The seismic anisotropy patterns become really similar between the three models at the higher amphibole content.

Early studies examining amphibole CPO revealed the presence of type I CPO amphiboles in various locations [92,93,94,95,96,97]. To gain insight into the evolution of the amphibole CPO, simple shear deformation experiments were conducted on amphibole at a high pressure of 1 GPa and temperatures ranging from 480 °C to 700 °C [43]. These experiments revealed that type I CPO amphibole predominates under low temperatures and varying stress conditions. Conversely, type II CPO is observed in the moderate temperature range (550–700 °C) and under high stress conditions. Type III CPO forms under high-temperature and low-stress conditions [43]. Research on the deformation fabric of hydrous minerals, including serpentine, chlorite, and amphibole, indicates that these minerals in natural rocks exhibit strong CPOs, which can generate significant parallel trench seismic anisotropy and result in longer S-wave delay times [40,43,91]. Detailed studies of the seismic velocity and anisotropy patterns of serpentine, chlorite, and amphibole reveal that the CPOs of hydrous minerals can explain the seismic anisotropy observed in strong parallel trenches at the stable plate-mantle interface, especially when the plate subduction angle is large (≥45°) [43].

6. Conclusions

The microstructures of olivine and amphibole, the primary mineral components of peridotite in the southern Mariana Trench, and their influence on seismic anisotropy are investigated. It is observed that olivine exhibits weak fabric strength, yet distinct deformation fabrics of types A/D, D, and AG can still be identified. The presence of various dislocation structures suggests that olivine undergoes intricate deformation processes at varying temperatures, leading to transformations in its original slip system. Additionally, the melted/fluid phase resulting from olivine deformation significantly impacts peridotite. This deformation precedes the melt/rock interaction, resulting in a strong melt/rock reaction under nearly static conditions. Moreover, the modal content of amphibole significantly alters the seismic anisotropy of peridotite. Although the geometry of the seismic anisotropy does not change much, an increase in the abundance of amphibole, whatever the CPO type, enhances the seismic anisotropy of peridotite, with type I amphibole being the most influential. Furthermore, the presence of type I fabric amphibole promotes the polarization direction of Vs1 parallel to the trench in the subduction zone. This phenomenon is also observed in other subduction zones. Consequently, the impact of amphibole on seismic anisotropy must be considered in mantle peridotite regions rich in amphibole.